DARPA Urban Challenge

The DARPA Urban Challenge

On November 3rd, 2007, the Defense Advanced Research Projects Agency (DARPA) hosted the Urban Challenge, an autonomous ground vehicle race at George Air Force Base in Victorville, CA. In order to complete the course, fully autonomous vehicles had to traverse 60 miles of urban and off-road environments in under 6 hours. Vehicles had to achieve speeds of up to 30 mph while negotiating live traffic from both human test drivers as well as the other robotic competitors. Each autonomous vehicle had to complete a series of simulated military supply missions requiring merging into moving traffic, navigating traffic circles, negotiating busy intersections, parking in marked lots, handling roadblocks, and avoiding obstacles. A landmark challenge in the field of unmanned systems, the DARPA Urban Challenge represents the state-of-the-art in mobile robotic ground vehicle technology.

“Yesterday was a historic day. For the first time ever, we saw robot-to-robot interaction at real speeds.”

- Tony Tether, Director of DARPA

Team VictorTango = TORC + VA Tech

Team VictorTango, a collaboration between TORC Technologies and Virginia Tech, was one of only 3 teams to successfully complete the challenge out of an original field of 89 international teams. Based out of Blacksburg, VA, team VictorTango consisted of university students, faculty, and TORC software engineers. TORC engineers served as lead software developers for team VictorTango, developing and testing the software needed to drive autonomously in an urban environment. Premiere sponsors of team VictorTango were Caterpillar Inc. and Ford Motor Company.

Odin: The Autonomous Hybrid Escape

Team VictorTango’s entry to the Urban Challenge was Odin, a 2005 Ford Hybrid Escape augmented with autonomous functionality. As a base drive-by-wire platform, Odin was safe, reliable, and capable of operating for extended periods of time without any human interaction. Odin took advantage of the factory hybrid power system as well as stock electronic control systems to minimize the need for additional hardware. As a result, Odin proved to have one of the cleanest, most seamless conversions to drive-by-wire across all competitors. TORC has commercialized this technology as the ByWire XGV™, presenting the same quality platform to system integrators, researchers, and developers worldwide.

Odin’s sensors gave it the ability to localize itself in the world and to perceive the surrounding environment. These sensors included 3 Multiplanar Laser Rangefinders (LIDAR), 4 Single-Plane LIDAR, 2 IEEE 1394 cameras, and an Inertial Navigation System (INS) composed of a Global Positioning System (GPS) and an Inertial Measurement Unit (IMU).

Odin’s main computing was supplied by a pair of HP servers equipped with two quad-core processors each. These servers ran TORC’s hybrid Deliberative-Reactive software architecture for autonomous ground vehicle control. This architecture implements optimal, search-based, planning techniques as well as reactive, behavior-based methods. Odin’s software was able to contend with complex situations in crowded, unpredictable environments while consistently obeying California state driving laws. This requires the sensing, classification, and tracking of both static and dynamic obstacles at high speeds. It also required the balancing of goals of dynamically changing importance, traversing the course as quickly as possible while remaining a safe and defensive driver.

Technology Transfer

TORC is focused on productizing key component technologies for unmanned systems, such as those developed to compete in the DARPA Urban Challenge. Through modular, commercial-off-the-shelf (COTS) components, TORC is maturing advanced technology pioneered by team VictorTango into hardened products and services. From hardware to software, TORC’s products shorten the development time of unmanned vehicles by system integrators, researchers, and developers.

The following products and services were extremely valuable in the DARPA Urban Challenge and have been successfully transferred to industry by TORC:

	SafeStop™ - A low-cost, multi-level, wireless emergency stop system, the SafeStop was used by many Urban Challenge teams and has been adopted as a safety standard in several commercial robotic applications.
	ByWire XGV™ - A Hybrid Escape drive-by-wire platform, the ByWire XGV makes Odin’s seamless conversion and superior design available to customers worldwide.
	AutonoNav™ - A ruggedized, compact computing module, the AutonoNav uses advanced Route Planning and Motion Planning software (similar to that used in the Urban Challenge) to rapidly convert ground and surface vehicle platforms from drive-by-wire to autonomous navigation.
	SimVironment™ - A distributed, scriptable, physics-based 3D simulation environment, SimVironment reduces software development times and allows for more complete and efficient testing while increasing safety and reducing cost.
	Autonomous Navigation System - Odin’s proven hybrid Deliberative/Reactive software architecture and interchangeable software modules can be easily adapted to custom mobile robotic applications, from open-pit mine automation to site surveillance and security.

Publications

For more information on Team VictorTango’s solution to the DARPA Urban Challenge, please refer to the following publications. This collection of Journal Papers and Masters Theses provide a much more detailed account of the research and technology behind Odin’s significant achievement.

A. Bacha et al., “Odin: Team VictorTango’s entry in the DARPA Urban Challenge,” Journal of Field Robotics, vol. 25, no. 8, Aug 2008.
J. Hurdus et al., “The VictorTango Architecture for Autonomous Navigation in the DARPA Urban Challenge,” Journal of Aerospace Computing, Information, and Communication, AIAA, 2009.
C. Reinholtz, “DARPA Urban Challenge Final Report,” 2007.
C. Reinholtz, “DARPA Urban Challenge Technical Paper,” 2007.
S. Cacciola, “Fusion of Laser Range-Finding and Computer Vision Data for Traffic Detection by Autonomous Vehicles,” Master’s Thesis, Virginia Tech, Blacksburg, VA, 2007.
J. Hurdus, “A Portable Approach to High-Level Behavioral Programming for Complex Autonomous Robot Applications,” Master’s Thesis, Virginia Tech, Blacksburg, VA, 2008.